Claude Enterprise gets usage controls

Anthropic added usage analytics and spend controls to Claude Enterprise on July 2. Admins can now see consumption broken down by user and group, with artifact creation counts, files edited, and which skills and connectors each team is using, alongside a cost line for each. An analytics chat interface accepts natural language queries against that data; an analytics API gives programmatic access to the same numbers.

The spend side adds model-level entitlements: admins set a default model for each user or group and can block access to more expensive models to prevent budget overruns. Threshold alerts fire at 75 percent and 90 percent of an organization’s monthly limit.

For operators managing Claude across a team, the controls address a real gap. Knowing that AI tools are being used is different from knowing how they are being used and what that costs. The AI governance playbook lays out the access-tier structure these entitlements are meant to enforce. The AI data privacy baseline covers what to check about where that usage data is stored and for how long.

The analytics API is the detail worth flagging for larger organizations. Direct programmatic access to consumption data means usage can feed a cost allocation workflow, a capacity forecast, or an audit log without manual export. That access should be treated like any external API key: scoped, rotated on a schedule, and logged.